Hormonal markers as noninvasive predictors of sperm retrieval in non-obstructive azoospermia

Abstract

Non-obstructive azoospermia (NOA) is one of the leading causes of male factor infertility, which results from impaired spermatogenesis. Currently, the sole feasible therapeutic option for men with NOA to father their biologic children is sperm retrieval by testicular sperm extraction (TESE) approaches followed by an intracytoplasmic sperm injection program. Nevertheless, the rate of sperm retrieval from NOA men following TESE has remained as low as 50%, leading to a significant number of unsuccessful TESE operations. Given that TESE is associated with multiple side effects, the prediction of TESE outcome preoperatively can abolish unnecessary operations and thereby prevent NOA patients from sustaining adverse side effects. As the process of spermatogenesis is under the regulation of hormones, the hormonal profile of serum and/or seminal plasma may contain useful information about spermatogenesis status and can potentially predict the chance of sperm retrieval from NOA patients. A large body of literature is available on the predictive capability of different serum and seminal plasma hormones such as FSH, LH, testosterone, inhibin B, AMH, estradiol, prolactin, and leptin in a stand-alone basis or combinational fashion with respect to the TESE outcome. The present review aimed to evaluate the potential of these hormonal markers as noninvasive predictors of sperm retrieval in men with NOA.

Introduction

Azoospermia is a male reproductive system disorder which is defined as the lack of spermatozoa in the two distinct semen specimens following centrifugation and microscopic analysis. The prevalence of azoospermia has been estimated at around 1% among men, accounting for approximately 10–15% of the men with male factor infertility [1]. According to the etiology, azoospermia is categorized into obstructive azoospermia (OA) and non-obstructive azoospermia (NOA). In the case of OA, despite the intact sperm production in the testes, obstruction in the reproductive tract prevents spermatozoa from entering the semen. On the contrary, NOA occurs as a consequence of impairment in the spermatogenesis [2, 3].

\Nowadays, the sole feasible therapeutic option for men with azoospermia to father their biologic children is testicular sperm extraction (TESE) followed by an intracytoplasmic sperm injection program [1]. Different techniques including fine needle aspiration (FNA), conventional testicular sperm extraction (cTESE), and microdissection TESE (mTESE) can be exploited for locating and retrieving spermatozoa from NOA subjects [4]. However, and irrespective of the technique used, clinically utilizable sperm are extracted from a maximum of approximately 50% of patients [5–7]. Given that TESE is associated with multiple side effects, especially hypogonadism [8], the prediction of TESE outcome preoperatively can abolish unnecessary operations and thereby prevent NOA patients from sustaining adverse side effects.

The process of spermatogenesis consists of a sequence of stepwise stages, which is under the regulation of various endocrine, paracrine, and autocrine hormonal stimuli. The hypothalamic-pituitary-gonadal axis (HPG axis) plays a major part in the regulation of sperm production and maturation. Accordingly, hypothalamic gonadotropin-releasing hormone (GnRH) stimulates the secretion of gonadotropins, i.e., follicle-stimulating hormone (FSH) and luteinizing hormone (LH), from the gonadotrophs of the anterior pituitary. These gonadotropins are major players in spermatogenesis development and maintenance, as they stimulate Sertoli cells and Leydig cells to support sperm production. Ultimately, gonadal steroid hormones and inhibin exert negative feedback control on GnRH secretion and thereby regulate gonadotropin levels and sustain the homeostasis of the HPG axis [9]. Besides the hormones of the HPG axis, other hormones such as prolactin and leptin were also found to participate in spermatogenesis directly by regulating testicular cells and/or indirectly by interacting with the HPG axis [10, 11].

Concerning the pivotal role of hormones, particularly those of the HPG axis, in the regulation of spermatogenesis, the hormonal profile of serum and/or seminal plasma may contain useful information about spermatogenesis status and can potentially predict the chance of testicular sperm retrieval from NOA patients. Hence, a large body of literature has evaluated the predictive capability of various hormones in either serum or seminal plasma of NOA men with respect to the TESE outcome. The present work aimed to review the latter literature and critically evaluate the value of different hormones as predictive biomarkers for sperm retrieval during TESE procedures.

Serum hormones and TESE outcome

Since blood is easily obtainable in comparison with testicular tissue and meanwhile is the main carrier of endocrine hormones contributing to the HPG axis and spermatogenesis regulation, the prediction of TESE outcome based on the hormonal profile of blood serum (or plasma) has attracted special attention in the last two decades. The present section elaborates on the predictive values of different serum hormones as noninvasive biomarkers of sperm retrieval in NOA patients (Table 1).

Table 1.

Predictive value of individual serum or seminal plasma hormones in terms of successful sperm retrieval by testicular sperm extraction

Study	Sample size	Statistical difference*	AUC	Cutoff value	Sensitivity (%)	Specificity (%)
Serum
Follicle-stimulating hormone (FSH)
[12]	100	Existed	NA	NA	NA	NA
[13]	62	Existed	0.77	18.3 IU/mL	58.8	84.4
[14]	85	Existed	0.96	9.9 IU/L	97.0	83.3
[15]	52	Not existed	NA	NA	NA	NA
[16]	191	Existed	0.656	15 IU/L	75.0	51.2
[17]	68§	Existed	0.830	22 IU/L	100	55
[18]	70§	Existed	NA	NA	NA	NA
[19]	36	Existed	0.889	19.08 IU/L	62	85
[20]	51§	NA	0.716	16 IU/L	74	74
[21]	792	Existed	NA	15 IU/mL†	NA	NA
[22]	139	Existed	NA	NA	NA	NA
[23]	280	Existed	0.656	20.5 IU/L	68.5	55.7
[24]	280	Existed	0.62	14.32 mIU/L	70.7	68.2
[25]	61	Existed	0.752	12.2 IU/L	71	75
[26]	131	Existed	NA	NA	NA	NA
[27]	1275	Not existed	NA	NA	NA	NA
[28]	12	Not existed	NA	22.03 IU/L	33.3	60.0
[29]	47	Not existed	NA	NA	NA	NA
[30]	145	Not existed	NA	NA	NA	NA
Luteinizing hormone (LH)
[12]	100	Not existed	NA	NA	NA	NA
[14]	85	Not existed	NA	NA	NA	NA
[16]	191	Existed	0.666	7.5 IU/L	63.1	63.9
[18]	70§	Not existed	NA	NA	NA	NA
[23]	280	Not existed	NA	NA	NA	NA
[24]	280	Not existed	NA	NA	NA	NA
[27]	1275	Not existed	NA	NA	NA	NA
[29]	47	Not existed	NA	NA	NA	NA
[30]	145	Not existed	NA	NA	NA	NA
Testosterone
[12]	100	Existed	NA	NA	NA	NA
[14]	85	Not existed	NA	NA	NA	NA
[16]	191	Existed	0.377	400 ng/mL	52.2	60.0
[18]	70§	Not existed	NA	NA	NA	NA
[19]	36	Existed	NA	NA	NA	NA
[23]	280	Not existed	NA	NA	NA	NA
[24]	280	Not existed	NA	NA	NA	NA
[27]	1275	Not existed	NA	NA	NA	NA
[29]	47	Not existed	NA	NA	NA	NA
[30]	145	Not existed	NA	NA	NA	NA
Inhibin B
[12]	100	Existed	NA	NA	NA	NA
[13]	62	Existed	0.85	34 pg/mL	70.6	95.6
[14]	85	Existed	0.84	39.8 pg/mL	72.2	95.5
[15]	52	Not existed	NA	NA	NA	NA
[16]	191	Not existed	NA	NA	NA	NA
[17]	68§	Existed	0.699	15 ng/L	62.5	81.5
[18]	70§	Not existed	NA	NA	NA	NA
[20]	51§	NA	0.610	14.5 pg/mL	100	29
[22]	139	Existed	0.694	27.5 pg/mL	60.7	78.1
[23]	280	Existed	0.683	27.5 pg/mL	60.4	69.5
[25]	61	Existed	0.851	67 pg/mL	80	79
[29]	47	Not existed	NA	NA	NA	NA
Anti-müllerian hormone (AMH)
[31]	24	Not existed	NA	NA	NA	NA
[20]	51§	NA	0.565	3.6 ng/mL	81	50
[29]	47	Existed	0.93	4.62 ng/mL	100	82
Estradiol
[12]	100	Not existed	NA	NA	NA	NA
[24]	280	Not existed	NA	NA	NA	NA
[29]	47	Not existed	NA	NA	NA	NA
Prolactin
[12]	100	Not existed	NA	NA	NA	NA
[16]	191	Not existed	NA	NA	NA	NA
[18]	70§	Not existed	NA	NA	NA	NA
[24]	280	Not existed	NA	NA	NA	NA
[29]	47	Not existed	NA	NA	NA	NA
Leptin
[24]	280	Not existed	NA	NA	NA	NA
Seminal plasma
Inhibin B
[13]	62	Existed	0.91	27.0 pg/mL	88.2	93.3
[17]	68§	Not existed	0.630	30 ng/L	66.6	66.6
[22]	139	Not existed	0.502	NA	NA	NA
[32]	78	Existed	NA	NA	NA	NA
Anti-müllerian hormone (AMH)
[33]	40	Not existed	NA	NA	NA	NA
[17]	68§	Not existed	0.535	1.4 pmol/L	77.7	66.4
[22]	139	Not existed	0.636	NA	NA	NA
[32]	78	Existed	NA	NA	NA	NA
Leptin
[24]	280	Existed	0.59	2.9 ng/mL	43.1	75.0

AUC area under the curve, NA not available or applicable

*Between groups of men with positive and negative sperm retrieval

^§Including patients with both obstructive and non-obstructive azoospermia

^†Surprisingly, higher values were significantly associated with greater chance of sperm recovery

Follicle-stimulating hormone

Serum FSH level has a good predictive value for active spermatogenesis, as it is useful to discriminate OA patients from NOA ones [34]. Consistently, several studies have found that the mean level of serum FSH was significantly lower in NOA patients with successful TESE outcome than those with unsuccessful outcome [12, 13, 19, 26]. Accordingly, the cutoff value of 9.9 mIU/mL for serum FSH was predictive of successful sperm retrieval with a sensitivity of 97% and a specificity of 83.3% in a report by Ziaee et al. [14]. Similarly, Bonarriba et al. [25] have introduced the cutoff point of 12.2 mIU/mL for serum FSH with sensitivity and specificity of 71% and 75%, respectively. Nevertheless, it is generally accepted that the serum FSH has a poor value for predicting the existence of sperm in the testes [27, 28, 30]. Indeed, cases with even serum FSH > 90 mIU/mL but positive TESE outcomes have been observed in previous studies [21]. Reaffirming the latter claim, a recent meta-analysis evaluated a total of 1261 individuals with NOA has elucidated that serum FSH is a poor predictor of successful sperm recovery by mTESE [35]. These controversies around the predictive value of FSH can be partially justified by the fact that concentration of serum FSH was found to correlate with the number of germ cells in the testes, but not the number of advanced differentiated spermatozoa [36]. Moreover, serum FSH level is merely representative of the overall spermatogenesis function and is unable to expose the foci of sperm existence in the testicular seminiferous tubules. Therefore, concerning the fact that mTESE which provides a precise examination of seminiferous tubules to find the spots with active spermatogenesis, the chance of sperm retrieval still exists in men with an elevated level of serum FSH [35, 37]. Taken together, although lower serum FSH levels correlate with higher sperm retrieval rate by TESE procedure, higher values do not eliminate the odds of successful sperm recovery, particularly via mTESE technique. Thus serum FSH does not appear to be a suitable biomarker for canceling the TESE procedure.

Luteinizing hormone

Most studies that investigated serum LH concentration with TESE outcome have found no significant difference in LH levels between patients with and without successful TESE outcome and hence did not consider any predictive value for serum LH level [12, 14, 18, 23, 27, 30]. By contrast, Cetinkaya et al. [16] have reported significantly lower levels of serum LH in men with positive sperm retrieval compared to those with negative sperm retrieval. They have also determined 7.5 mIU/mL as a cutoff point for serum LH and have introduced LH as one of the positive predictive biomarkers for the sperm retrieval rate. Nevertheless, the sensitivity and specificity of LH cutoff value in the latter report was 63.1% and 63.9%, respectively, which are much lower than acceptable values for a biomarker with the potential of clinical application. There is a consensus in the literature that the rate of sperm retrieval through the TESE procedure is not accurately predictable based on serum LH level.

Testosterone

Few studies found significantly different levels of testosterone in serum between groups of men with positive and negative TESE outcome and took account of serum testosterone level as a potential biomarker for predicting TESE success [12, 16]. Accordingly, the significantly higher levels of both total and free testosterone were detected in the serum of NOA patients with successful sperm retrieval compared to those with unsuccessful sperm recovery in reports by Tsujimura et al. [12]. Additionally, owing to possess the highest regression coefficient in the multivariate regression model, serum total testosterone was considered the most effective predictive factor for successful sperm retrieval in the latter study. Similarly, Cetinkaya et al. [16] observed that men with positive TESE outcomes have significantly higher levels of both total and free testosterone in their serum than those with negative TESE outcome. In addition, they obtained cutoff points of 400 ng/mL for total testosterone and 11 pg/mL for free testosterone with predictive accuracies of 37.7% and 35.5%, respectively. By contrast, the majority of studies analyzed both total and free levels of serum testosterone in the aforementioned groups of patients did not corroborate this statistically significant difference between the groups and therefore have concluded that serum testosterone lacks any value in predicting sperm retrieval rate by TESE operation [14, 18, 19, 23, 27, 30]. Moreover, in reports which proposed testosterone as a predictive biomarker, serum testosterone level either acts as a part of a multivariate model or lacks sufficient sensitivity and specificity required for clinical application. Therefore, serum testosterone level does not seem an appropriate hallmark of sperm retrieval before the TESE operation.

Inhibin B

The predictive value of serum inhibin B in terms of sperm retrieval is confusingly contradictory in the literature. On the one hand, several studies found no predictive value for inhibin B due to the lack of significant difference in serum inhibin B levels between TESE-positive and -negative men [15, 16, 18, 29]. On the other hand, multiple pieces of research indicated statistically different values of serum inhibin B between men with successful sperm retrieval and those with unsuccessful sperm retrieval and thus recommended inhibin B as a valuable biomarker for forecasting TESE outcome [12–14, 22, 23, 25]. For instance, by evaluating 280 NOA patients, Boitrelle et al. [23] have obtained 27.5 pg/mL as a cutoff point for serum inhibin B with sensitivity and specificity of 60.4% and 69.5%, respectively. Similarly, the cutoff value for inhibin B was obtained 27.5 pg/mL in the report by Mitchell et al. [22] with sensitivity and specificity of 60.7% and 78.1%, respectively. Nevertheless, there are considerable variations between the calculated cut-off points in the different studies, as while it was 27.5 pg/mL in the latter studies, the cutoff values of 39.8 and 67 pg/mL were obtained in the reports by the Ziaee et al. [14] and Bonarriba et al. [25], respectively. The other limitation with this respect is that reported cutoff values do not possess sufficient sensitivity and specificity to forecast the chance of sperm recovery with acceptable accuracy. Indeed, there is a notable proportion of patients with low serum inhibin B that in whom TESE results in successful sperm retrieval [25]. Overall, there is an agreement in the literature that serum inhibin B cannot be single-handedly supposed to be a perfect predictive biomarker for sperm retrieval rate to avoid performing unnecessary TESE operations.

Anti-müllerian hormone

Most of the studies that investigated the serum values of AMH in relation to the spermatogenesis status of men did not establish a significant association. For instance, although serum AMH levels in fertile men were statistically higher than subfertile ones, there was no significant difference among various causes of subfertility in terms of serum AMH in the report by Goulis et al. [38], indicating that measurement of serum AMH concentration does not appear to be useful for differentiating etiology of subfertility. Similarly, serum AMH levels in normozoospermic men did not exhibit significant differences with that of oligospermic [39] and azoospermic [40] patients, suggesting the low value of serum AMH in the prediction of impaired spermatogenesis. Consistently, it is also reported that serum AMH concentration is not able to provide predictive information on the probability of sperm retrieval in NOA patients who underwent FNA [20] or TESE [31] procedures. On the contrary, by studying 47 white-Caucasian men with idiopathic NOA, Alfano et al. [29] recently demonstrated that serum levels of AMH act as an independent predictor of successful sperm recovery by mTESE with a predictive accuracy of 93%. In addition, by calculating the predictive cutoff value of 4.62 ng/mL for serum AMH, they achieved 100% sensitivity and 82% specificity, with a positive predictive value of 79% and a negative predictive value of 100% for successful sperm recovery [29]. These findings suggest that serum AMH can emerge as a biomarker with a considerable clinical significance, which its utilization in the practice can decrease the rate of unnecessary surgical interventions and allow to promote TESE outcome in the patients with a high chance of sperm retrieval. However, the results of the latter study, notwithstanding their salience, should be interpreted cautiously due to the existence of significant limitations in that study: Firstly, relatively small sample size and secondly, the exclusion of diverse groups of men with respect to NOA etiology makes it difficult to generalize its findings to the population of NOA patients. In fact, it has been indicated that men with various causes of NOA such as AZFa or AZFb microdeletion, and Klinefelter syndrome have significantly different chances of successful sperm recovery by TESE [27]. Therefore, the inclusion of men with various etiologies of NOA in the study of Alfano et al. could alter the sperm retrieval rate by mTESE surgery and thereby influence its association with the concentration of circulating AMH. Taken together, the predictive value of serum AMH with regard to TESE outcome has not achieved consensus so far and it remains to be determined in the large cohort of the population whether serum AMH levels are associated with sperm retrieval rate by sperm extraction techniques.

Estradiol

There are relatively few studies that investigated serum estradiol in relation to the success of sperm recovery during the TESE procedure, which most of them found that positive and negative groups are comparable in terms of serum estradiol concentration [12, 24, 29]. Similarly, Basar et al. [41] observed no statistical difference for serum estradiol levels between successful and failed sperm retrieval groups in NOA men who underwent mTESE for the first time. However, they noticed significantly higher levels of serum estradiol in the patients with the repeated history of mTESE compared to the primary group, indicating a negative correlation between serum estradiol levels and the number of previous mTESE implementation. These findings suggest the potential value of serum estradiol levels in predicting the minimum number of attempts required for a positive outcome by sperm retrieval techniques. However, it should be considered whether repeated TESE implementations influence the testicular production of estradiol.

Prolactin

Several studies have evaluated serum prolactin prior to TESE implementation in terms of successful sperm recovery. Despite the previous hormonal biomarkers which their discriminative values were inconsistent in the literature to some degree, neither of these studies found a statistical difference in the serum levels of prolactin between NOA patients with favorable and unfavorable outcomes [12, 16, 18, 24, 29]. Therefore, it is evident that serum prolactin lacks any predictive remark for the chance of sperm recovery by the TESE procedure.

Leptin

To the best of our knowledge, there is one study in PubMed that has investigated the serum levels of leptin in NOA patients in relation to the chance of sperm retrieval from the testes. In this study, by evaluating 280 NOA patients who underwent TESE surgery, Ma et al. [24] found no predictive value for serum leptin in terms of TESE outcome due to the lack of significant difference in the serum levels of leptin between men with successful and unsuccessful sperm recovery. However, more studies on the large and various cohort of NOA patients are required to corroborate whether serum leptin is valuable in differentiating superior candidates for TESE implementation from the men with a greatly poor chance of sperm retrieval by TESE.

Seminal hormones and TESE outcome

In comparison with serum, the profile of seminal plasma hormones, particularly those which are produced and secreted by testicular cells, may be more relevant to spermatogenesis status and consequently predict the existence of spermatozoa in the testicular tissue more accurately. The present section addresses the predictive values of seminal plasma hormones to predict the outcome of TESE in men with NOA (Table 1).

Inhibin B

Because of predominant secretion from the apical pole of Sertoli cells into seminiferous tubules, inhibin B levels in the seminal plasma may be approximately 10-fold higher than serum values [22]. There are few studies concerning the predictive value of seminal plasma inhibin B for TESE outcome, in which existent results are not consistent. For instance, Nagata et al. [13] have observed statistically higher levels of seminal inhibin B in the positive TESE group compared to the negative group. Further analyses by the authors yielded 27.0 pg/mL as a discriminative cutoff value for seminal inhibin B, which could predict the success of sperm retrieval with sensitivity and specificity of 88.2% and 93.3%, respectively [13]. By contrast, some other evidence indicated no significant differences between men with successful and failed sperm recovery in terms of seminal plasma inhibin B [17, 22]. Furthermore, seminal inhibin B has found to be even unable to identify the groups of patients with or without germ cells in the testicular biopsy specimens [22]. Irrespective of predictive accuracy, the large variability of seminal plasma inhibin B levels also restricts its clinical applicability as a predictor of TESE outcome, as Mitchell et al. [22] could not detect inhibin B in the seminal plasma of 96 cases out of 139 patients with NOA. Taken together, seminal plasma inhibin B does not appear to be an appropriate criterion for predicting sperm retrieval during TESE.

Anti-müllerian hormone

AMH is another Sertoli cells-derived hormone that exists in high levels within the seminal plasma due to preferential apical efflux toward seminiferous lumen [22]. It can be declared that the literature data concerning the prediction of surgical sperm recovery according to seminal levels of AMH has achieved consensus. Accordingly, seminal levels of AMH have been found to predict the presence of neither spermatozoa nor germ cells in the testicular tissue of NOA cases underwent TESE surgery [22]. Likewise, similar studies did not exhibit discriminative remarks for seminal AMH in terms of successful TESE outcomes [17, 33]. Moreover, seminal AMH values range widely and were undetectable in a remarkable proportion of patients, which precludes its applicability in clinical practice [17, 22, 33]. Overall, seminal AMH cannot be considered a prognostic biomarker for positive sperm extraction.

Leptin

Leptin levels in the seminal fluid have been found to correlate with poor sperm motility and increased sperm DNA fragmentation [42], suggesting that high concentrations of leptin in the seminal plasma can adversely influence the spermatogenesis process. Consistently, NOA patients with failed sperm recovery exhibited significantly higher levels of seminal leptin than men with successful TESE outcomes in a report by Ma et al. [24]. The predictive accuracy of seminal leptin was derived 0.59, with the best discriminative value of 2.9 ng/mL (sensitivity 43.1%, specificity 75.0). Given the improvement in the predictive accuracy of sperm extraction following a combination of seminal leptin with other routine predictors (e.g., serum FSH, testicular volume), the authors introduced seminal leptin as an assistant predictor of TESE outcome. However, additional studies are required to verify these results.

Hormonal markers as components of prediction models

A prediction model is a statistical formula consisting of a number of predictor variables (demographic information, laboratory tests, and disorder attribute) which is built to estimate the odds of a specific future outcome (e.g., disease, incident, and complications) in individual patients with certain status of health. Preferably, these models provide an approximation of absolute probability but may produce relative probability or probability score. Over the last years, prediction models have attracted much attention in the development of prognostic tools for various medical circumstances [43]. Consistently, several studies have reported multivariable prediction models for estimating the chance of sperm retrieval by various surgical methods, including TESE, in men suffering from NOA. Importantly, hormonal markers, without exception, are of main components of these prediction models (Table 2). For instance, Tsujimura et al. [12] have proposed a multivariable prognostic model consisting of serum FSH, total testosterone, and inhibin B for estimating the odds of positive mTESE outcome in a cohort of 100 NOA patients. The model displayed an overall accuracy of 0.76 with a sensitivity of 71% and a specificity of 71.4% at the cutoff point of 15.7%, wherein FSH was a negative predictor and both testosterone and inhibin B were positive predictors of successful sperm recovery. In a similar study, Boitrelle et al. [23] used serum levels of FSH and inhibin B along with total testicular volume as independent predictors to calculate a discriminative score for TESE outcome in a cohort of 280 NOA individuals. Accordingly, the model resulted in a moderate power (0.663) with a fair predictive specificity (82.4%) but poor sensitivity (53%) at the cutoff value of 18.5, in which successful sperm recovery was correlated with lower concentrations of FSH and higher values of both serum inhibin B and testicular volume. Regarding the ease of application and low invasiveness of testicular sperm aspiration (TESA) in comparison with TESE procedures, Majzoub et al. [46] recently attempted to develop a predictive model for sperm retrieval by TESA in a population of 297 NOA patients. Of multiple parameters evaluated such as testis size, serum hormones, chromosomal karyotype, and Y chromosome deletions, only testis size and serum FSH appeared as significant predictors and the model resulting from them could predict TESA outcome with an accuracy of 0.742, sensitivity of 73.9%, and specificity of 69.9%. Additionally, the authors noted that the values of > 7.75 mL and < 8.5 IU/L for testis size and FSH, respectively, were associated with more than 5-fold increment in the probability of successful sperm aspiration, suggesting that TESE can be replaced with less traumatic TESA in a certain subpopulation of NOA men displaying mean testicular volume > 7.75 mL and serum FSH < 8.5 IU/L concomitantly.

Table 2.

Multivariable prediction models for successful sperm retrieval by testicular sperm extraction

Reference	Sample size	Model or predictor variables	AUC	Cutoff value	Sensitivity (%)	Specificity (%)
[12]	100	P = 1 / 1 + exp (−β) β = −5.201 + (0.048 × FSH) + (0.449 × TT) + (0.021 × INH B)	0.76	15.7%	71.0	71.4
[17]	68§	β = 4.6815 + (0.2174 × Seminal INH B) + (1.3418 × Seminal AMH) – (0.8099 × FSH)	0.985	1.358	100	80
[20]	51§	β = INH B × AMH × Volume of the large testis / FSH	0.669	49	89	50
[23]	280	Score = 915 × FSH / Total testicular volume × INH B	0.663	18.5	53.0	82.4
[24]	280	Testicular volume; semen volume, seminal pH, seminal alpha-glucosidase and fructose; serum FSH, LH, TT, prolactin, estradiol; serum and seminal leptin†	0.832	NA	NA	NA
[44]	1371	P = 1 / 1 + exp (−β) β = −1.009 + (Male age × 0.058) + (LH – 0.115) + (LH2 × 0.001) + (FSH – 0.019) + (TT × 0.034) + (AZFc deletion – 1.480) + (Idiopathic NOA – 0.855)	0.69	NA	NA	NA
[45]	597	P = 1 / 1 + exp (−β) β = 0.003 + (0.124 × FSH) + (−1.448 × Age 35 years or older) + (−0.651 × Testicular volume 12–14 mL) + (−1.382 × Testicular volume 15 mL or greater)	0.823	64.61%	67.7%	86.3%
[46]	297	P = 1 / 1 + exp (−β) β = −1.069 + (−0.081 × FSH) + (0.079 × average testis size)	0.742	23.13%	73.9%	69.9%

β = regression coefficient. AMH anti-müllerian hormone, AUC area under the curve, FSH follicle-stimulating hormone, INH B inhibin B, LH luteinizing hormone, NA not available or applicable, NOA non-obstructive azoospermia, TT total testosterone

^§Including patients with both obstructive and non-obstructive azoospermia

^†Artificial neural network model

In addition to the above-mentioned findings, other studies employed seminal hormones besides those of serum as independent predictors of multivariable models for sperm retrieval rate by TESE procedure. In this respect, Duvilla et al. [17] developed a prediction model by integrating serum FSH, seminal inhibin B, and seminal AMH values from a group of 68 men with azoospermia. Interestingly, with a predictive power of 0.985, a sensitivity of 100%, and a specificity of 80% at a cutoff point of 1.358, the model demonstrated extremely outstanding capability to estimate successful sperm recovery from patients. By exploiting artificial neural networks among a population of 280 NOA individuals, Ma et al. [24] also built several multivariable models through the different combinations of (1) testicular volume, (2) semen volume, seminal pH, seminal fructose and alpha-glucosidase, (3) serum values of FSH, LH, total testosterone, prolactin, estradiol, (4) serum and seminal concentrations of leptin. Accordingly, all models were superior to serum FSH alone in terms of differentiating between sperm positive and sperm negative categories, among which a model derived from the combination of all parameters displayed the best predictive accuracy (0.832).

The establishment of reliable models for clinical utility requires a first-rate set of data from a large sample size as well as validation in a separate set of data from another independent population [43]. Given that the aforementioned models were constructed as a result of investigating a relatively limited number of azoospermic men and were not also validated in an external cohort of patients, they are of restricted practical value through various centers. To provide an independently validated prediction model for the chance of sperm retrieval, Cissen et al. [44] carried out a large countrywide study including 1371 distinct TESE operations on NOA men. According to multivariable analyses, the male age along with serum concentration of testosterone was positively and serum levels of FSH and LH together with the diagnosis of idiopathic NOA and AZFc deletion were negatively associated with successful sperm recovery. The model resulting from the combination of these independent predictors demonstrated overall predictive accuracies of 0.69 and 0.65 in developmental and validation datasets, respectively, suggesting moderate efficiency of the model in terms of the discrimination between men with good and poor prognosis of TESE outcome. Analogous to this study, Ma et al. [45] built and validated a multivariable prediction model for the risk of unsuccessful sperm recovery by FNA in a total of 597 men with NOA. Accordingly, higher serum FSH levels, lower testicular volume, and younger ages were independently predictive of sperm retrieval failure. In the developmental phase, the construction of a multivariable model using serum FSH level, testicular volume, and male age resulted in a cumulative power of 0.823 together with sensitivity and specificity of 67.7% and 86.3%, respectively, at 64.61% as the best cutoff value. The predictive power of the model was verified in the external validation phase, wherein the model was capable of predicting unsuccessful FNA cycles accurately in approximately 85% of cases. Altogether, multivariable prediction models offer a promising prospect for predicting the success of intra-testicular sperm retrieval in NOA men. Furthermore, the substitution of seminal hormones, especially those with Sertoli cell origin, seems to improve the predictive accuracy of these models, although it remains to be validated by future research.

Conclusion and prospect

In the present review, the potential of various serum and seminal plasma hormones as markers for predicting the success of sperm acquisition using TESE approaches have been critically elaborated. Accordingly, on a stand-alone basis, there is no consensual hormonal marker that is capable of discriminating between successful and unsuccessful sperm retrieval outcomes with satisfactory accuracy. Furthermore, while most of the prediction models constructed from the combination of various hormonal and non-hormonal factors do not present perfect predictive accuracy, available data suggest that the combination of serum FSH, seminal inhibin B, and seminal AMH may provide a powerful model to predict the chance of sperm retrieval. However, further research in large sample sizes especially on NOA men are required to confirm the predictive efficiency of these hormones both individually and in combination.

In addition to hormones, serum and seminal plasma contains a variety of other biological compounds including genomic DNA, tissue-exclusive RNAs, proteins, and intermediary metabolites, which can serve as noninvasive molecular biomarkers for a wide range of pathological conditions related to male fertility [47]. In this respect, microdeletions in specific sequences of the Y chromosome, namely AZFa, AZFb, and AZFc, have emerged as the first and up to now the most conclusive predictors of sperm retrieval rate. In fact, it has been indicated that the chance of sperm recovery from men with AZFa and AZFb deletions almost tends to zero, while AZFc deletion may be associated with up to 75% the probability of testicular sperm existence [48]. Due to providing protected and enriched sources of biological molecules, exosomal vesicles have recently received particular attention as the targets of biomarker discovery [49]. For example, the efficiency of exosomal microRNAs (miRNAs), especially miR-539-5p and the miR-941, in seminal plasma to predict the existence of focal spermatogenesis sites in a group of men with severe spermatogenesis impairment has been demonstrated [50]. Likewise, a seminal panel of exosomal long non-coding RNAs (lncRNAs) elicited from NOA patients displayed an outstanding predictive value in terms of sperm retrieval rate according to a recent report. Interestingly, the aforementioned lncRNA panel had remarkably higher predictive precision for the recovery of sperm from the testicular biopsies of NOA men in comparison with two different hormone-based models [51]. Moreover, there are numerous studies that evaluated seminal plasma proteome in azoospermic patients, although the majority of them have focused on the diagnosis and differentiation of azoospermia subclasses, i.e., OA and NOA, rather than the success of TESE outcome. Indeed, it has been indicated that the incidence of NOA is associated with the differential expression of multiple seminal plasma proteins such as underexpression of prolactin-inducible protein (PIP) and overexpression of gamma-glutamyltransferase 7 (GGT7), sorbitol dehydrogenase (SORD), epididymis-expressed extracellular matrix protein 1 (ECM1), and the testis-expressed sequence 101 (TEX101), suggesting the potential of these proteins, either alone or together, as noninvasive biomarkers for differentiating NOA from OA. Importantly, among these proteomic biomarkers, TEX101 also showed the capability to provide prognostic information on the histopathological classification of NOA patients, which enables it to predict the chance of sperm acquisition at least to some degree [52]. Finally, some preliminary attempts have been made to identify biomarkers for TESE outcome in NOA men based on the metabolomic profile of seminal plasma, which corroborated the possible utility of metabolomics analysis as a sensitive and specific strategy for discriminating NOA men with positive TESE outcome from those with negative outcome presurgically [53, 54].

Despite growing interest in the exploration of molecular biomarkers and their promising potential to act as effective tools for predicting the chance of sperm retrieval in NOA, there are serious limitations in relation to them at present. Firstly, the predictive value of these biomarkers has predominantly been observed in only one study with a small number of NOA men and consequently is inconclusive and questionable. In support of this, while miR-539-5p emerged as a good predictor of successful sperm recovery in one study [50], it exhibited no predictive value regarding TESE outcome in another study [51]. Hence, it seems reasonable to assume that further examining the predictive potential of the aforementioned molecular biomarkers can provoke similar controversy observed for hormonal markers. Secondly, and in contrast to hormones, not only abundance and stability of molecular biomarkers in seminal plasma are challenging but also their quantification methods are sophisticated, labor-intensive, and expensive, which hinders the process of their transition from discovery to clinical application. Therefore, even with the advent of transcriptomics-, proteomics-, and metabolomics-based biomarkers, due to abundance, stability, ease of measurement, and pivotal role in spermatogenesis, hormones of either serum or seminal plasma still have the potential to be a part of future investigations for predicting sperm retrieval rate from men with NOA.

Declarations

Conflict of interest

The authors declare no competing interests.